[Help] Using Pretrained Model

[Hussain Abdullah]

Thanks for sharing the model. I just want to test the pertained model that you provided. Do I still need to download the image data (SynthText/MjSynth) if I'm using the pretrained model? And If not then how can I get run the pertained model on testing datasets like cute80, idcars etc. I have already downloaded the datasets (cute80, idcar2013, idcar2015, iiit5k, svt, svtp) and their respective npz files. How can I run the evaluation on these datasets?

[Christian Bartz]

Hi,
First, you'll need to tell our code where your evaluation files are in the config file.
Then, need to unpack the model, place the files in any dir (e.g. trained_models/pre_trained) and then you can run the script run_eval_on_all_datasets.py, as described in the Evaluation section of our README.

That should be it.

[Hussain Abdullah]

I don't have cuda compatible GPU... I'm trying to run the model on MacBook Pro 2018... How can I run the pertained model on CPU... I'm getting this runtime error.

RuntimeError: CUDA environment is not correctly set up
(see https://github.com/chainer/chainer#installation).No module named 'cupy'

[Christian Bartz]

It should work on CPU. Could you provide the exact command you used and also the complete output of the script?

[Hussain Abdullah]

I ran the following command (Pretrained model files are in model folder in project's root directory)

 python run_eval_on_all_datasets.py config.cfg 0 -b 1 --snapshot-dir model --render

And the Error I got is this:

Testing cute80
stripping non alpha
Traceback (most recent call last):
  File "evaluate.py", line 310, in <module>
    evaluator = Evaluator(args)
  File "evaluate.py", line 105, in __init__
    self.localizer.to_device(args.gpu)
  File "/opt/anaconda3/envs/env/lib/python3.8/site-packages/chainer/device_resident.py", line 196, in to_device
    device = chainer.get_device(device)
  File "/opt/anaconda3/envs/env/lib/python3.8/site-packages/chainer/backend.py", line 149, in get_device
    return _get_device_cupy_or_numpy(int_device_spec)
  File "/opt/anaconda3/envs/env/lib/python3.8/site-packages/chainer/backend.py", line 188, in _get_device_cupy_or_numpy
    return cuda.GpuDevice.from_device_id(device_spec)
  File "/opt/anaconda3/envs/env/lib/python3.8/site-packages/chainer/backends/cuda.py", line 228, in from_device_id
    check_cuda_available()
  File "/opt/anaconda3/envs/env/lib/python3.8/site-packages/chainer/backends/cuda.py", line 142, in check_cuda_available
    raise RuntimeError(msg)
RuntimeError: CUDA environment is not correctly set up
(see https://github.com/chainer/chainer#installation).No module named 'cupy'
Traceback (most recent call last):
  File "run_eval_on_all_datasets.py", line 114, in <module>
    subprocess.run([command, file] + process_args, check=True)
  File "/opt/anaconda3/envs/env/lib/python3.8/subprocess.py", line 512, in run
    raise CalledProcessError(retcode, process.args,
subprocess.CalledProcessError: Command '['python', 'evaluate.py', '--gpu', '0', 'Eval_Datasets/CUTE80/gt.npz', 'model', 'LSTMTextLocalizer_', '--recognizer-name', 'TransformerTextRecognizer_', '--char-map', 'train_utils/char-map-bos.json', '--results-path', 'cute80_eval_results.json', '--dataset-name', 'cute80', '--strip-non-alpha', '--save-predictions', '--do-not-cut-bboxes', '--render-all-results', '-b', '1']' returned non-zero exit status 1.

[Christian Bartz]

Ah yes I see.

The script expects you to supply a GPU id (the argument following the config file). You supplied it with 0 so the script thinks, it should take the GPU with ID 0. If you want to use the CPU just supply -1 and it should work. If not, I might need to adapt the code, which is quite simply though.

Just exchange this by this statement: parser.add_argument("--gpu", default='cpu', help="GPU Id to use") and it should work.

[Hussain Abdullah]

It worked.... Thanks....